12.45K
total weekly downloads
15 packages
@strixgov/capabilities-claude-code
Pre-classified capability registry for Claude Code's built-in tools (Read, Write, Edit, Bash, Glob, Grep, Task, NotebookEdit, WebFetch, WebSearch). Drop-in starter for @strixgov/tool-gateway.
@strixgov/capabilities-mcp-common
Pre-classified capability registry for popular MCP servers (Slack, GitHub, Linear, Notion, Filesystem, Postgres, Email/SMTP/Gmail). Drop-in starter for @strixgov/tool-gateway.
@strixgov/execution-authority
The external SDK for the Strix Execution Authority Boundary: ask Strix for bounded authority over one exact proposed action, redeem it immediately before the handler runs, report the outcome, and independently verify the signed evidence. Zero shared code
@strixgov/governed-action
Wrap one consequential mutation — a REST call, a GitHub merge, any async operation — in a governed action that produces an independently verifiable signed receipt.
@strixgov/governed-egress
Strix Governed AI Egress — data-release control immediately before external AI execution. Detects sensitive data, de-identifies by consistent tokenization, re-scans, and binds a single-use egress authorization to the exact sanitized artifact so the artifa
@strixgov/guard
The seatbelt for MCP agents — one command wraps your MCP servers with Strix governance: writes wait for an approval, reads pass, every action produces a signed receipt.
@strixgov/jason
Jason Executive Orchestrator — converts executive intent into governed, stakeholder-reviewed work orders. Not an execution or authorization system.
@strixgov/mcp-adapter
One-call governance for any MCP server. Wraps every tool.call() with policy evaluation, signed execution receipts, and an optional approval gate — no changes to your tool implementations required.
@strixgov/mcp-proxy
Strix MCP Proxy — the distribution wedge for governed agent tool execution. Point your MCP client at the proxy instead of the underlying server; every callTool is intercepted, evaluated against policy, and emits a signed Ed25519 receipt. Designed so Mode
@strixgov/mcp-token-validator
Independent validator for Strix execution_authorization_v1 tokens. Drop into your first-party MCP server, network proxy, or credential-broker wrap to enforce Mode 3 (capability-enforced) governance — no Strix runtime dependency, no network calls. Verifies
@strixgov/rcm-reference
Canonical governed-agent reference implementation for healthcare revenue-cycle management (RCM). The same prior-authorization agent runs the same task ungoverned (policy + guardrails only) and governed by a Strix execution gate, across six real-world fail
@strixgov/sdk
Strix open proof surface — verify governed-action receipts, decision tokens, and actor attestations against the public JWKS without trusting Strix. Canonicalization, schemas, receipt signing, JWKS resolution, and verification badge. (The control layer — p
@strixgov/swarm-adapter
Bring your swarm. Strix governs it. A framework-neutral bridge that turns a real multi-agent run (Claude Agent SDK / CrewAI / LangGraph / OpenAI Swarm) into signed delegations + governed tool calls + independently-verifiable proof.
@strixgov/tool-gateway
Local-first governed tool execution gateway for AI agents. Read/write classification, policy evaluation, signed execution receipts, multi-key rotation with cross-signed chain snapshots, per-capability rate limits, threshold-based escalation, file/webhook
@strixgov/verifier
Independent verification of Strix governance evidence, approval artifacts, linked attestations, tool-gateway authorization receipts, signed execution outcomes, chain snapshots, and connected-mode wire envelopes. Offline-capable. No Strix account required.